如何加快FULL OUTER JOIN的速度

Question

我生成两个新旧表，它们包含相同的行集。有什么方法可以加快他们的加入速度？

WITH new as (
    SELECT uid, segid, true AS b
    FROM all_segments
    ),

    old as (
    select uid, segid, bool_or(status) as b
    from users_segment_history_current
    group by uid, segid
    )

    select uid, segid, now(), coalesce(new.b, false), false , now() - ('-420 min')::interval, old.b as old_b
    from new full outer join old using (uid, segid)
    where coalesce(new.b,false) is distinct from old.b;

解释分析：

QUERY PLAN                                          
Merge Full Join  (cost=52202.14..55745.83 rows=68964 width=28) (actual time=1565.272..1918.952 rows=219185 loops=1)                                         
  Merge Cond: ((new.uid = old.uid) AND (new.segid = old.segid))                                         
  Filter: (COALESCE(new.b, false) IS DISTINCT FROM old.b)                                           
  Rows Removed by Filter: 271427